By dhushon on May 07, 2006
An article, published on 5/5 at Fortune.com seems to be a harbinger of something either interesting or sinister happening in the Internet/2 realm. Here we have Google which makes the majority (like 98+%) of their revenue through advertising moving to take on a large role in information meta-indexing and even data management. And then we have Microsoft, who sells software with 95% of their revenue from the sale of “old-economy” shrink wrapped software, also moving to use data management & storage as a mechanism to build stickyness into their service economy (and I might add to execute on a defensive strategy against Google and the Free/Open-Software movements).
If you look at the cost of operating a data-center, they tend to get organized into a couple of categories:
- Facilities expenses: dominated by power and bandwidth
- Operational expense: dominated by labor
- Capital expense: increasingly dominated by data management systems but also by computers
- Software expense (sometimes capitalized): dominated again by in typical order: data management systems, vertical integration software, and infrastructure middleware
There are different strategies employed by different kinds of hosting, hardware, infrastructure software vendors to minimize these costs either within category or by trying to integrate capabilities and therefore uses systemic engineering approaches to minimize total cost. If one looks at Google, for example, they run their own data centers which allows them to maximize the interaction between compute/storage density and the facility costs (they control all the variables in 1/2/3 above - the typical IT charges), as well as rolling their own OS image, middleware software, and even File Systems (GFS) to manage 2/3/4. Microsoft, on the other hand, with their core business in engineering hyper-integration of 4 to reduce costs, needs to understand the impact of doing 1,2,3 on their existing channel and determine whether they can afford to do the focussed engineering to do the feature reduced / reliability enhanced engineering around their integrated software stack to achieve 1,2,3.
What is interesting to me is the yet undefined intersection of business data stores, and the online shared data sets (see Microsoft's Map the world in real-time initiative). This is to say the information economy is built around the premise that those who can can have the best information (or process information the most accurately and efficiently) will have more value than other category players. So if we start having these data centers (note the term data centers is very accurate here), built through advertising $'s, at what point do the “owners” of this data, or potentially more importantly the data's relationships begin to control a macro-economy, able to exploit this ownership to incredible, even monopolistic possibility.
I have to say, and it should come as no surprise, that the notion of utility data centers which aggregate demand against a shared physical plant does offer the possibility of democratized computing. Utility Data Centers effectively level the playing field between those who have capital and those that don't. By creating an open-marketplace in which the efficient use of assets, and the “value” of the services that one develops enables a small player to expose their innovation and compete successfully against the large players. With the recent moves by Microsoft and Google, one has to wonder if this is yet another emerging market where people fail to recognize the potential for monopoly abuse until it's too late, until the critical data needed by other “information economy players” becomes owned by a select few, and access to this data becomes too costly to compete.